4. Why DBMS in Cloud?
Database Management Systems as a cloud service are engineered to run as a scalable,
elastic service available on a cloud infrastructure. These DBMS are available only as a
cloud offering and are not necessarily relational. For example, Microsoft’s SQL Azure is
fully relational DBMS, while Microsoft’s SQL services, Amazon’s simpleDB and
Google’s Big Table are not relational and have different persistence models. Cloud-based
DBMS services are provided in a multi-tenancy environment with elastic resources
allocation, for use in simple to complex transactions. DBMS as a cloud service excludes
those DBMS that will run on the cloud infrastructure, but are not purpose-built as a cloud
service. Most of the currently available DBMS engines will run on cloud infrastructure,
but are not specifically engineered to take advantage of the cloud. This differentiation is
the reason for the change in name from “DBMS in the Cloud” to “DBMS as a cloud
Service”; running on cloud infrastructure does not define a DBMS as a cloud service [2].
All currently available cloud DBMS are relatively new. SQL azure, the only fully
relational DBMS available, began full production at the beginning of 2012 and still has
some size limitations; Microsoft plans to reduce, and eventually lift, these restrictions.
Today, DBMS as a cloud service are used primarily for development and testing of
applications- where database sizes are small and issues of security and collocation with
multiple users are not concern. One big advantages of cloud DBMS is their elasticity: the
more you use, the more you pay; the less you use, the less you pay [2].
Initially, cloud DBMSs will have an impact for vendors desiring a less expensive
platform for development. As cloud infrastructure with DBMSs gains maturity especially
in scalability, reliability and security, cloud implementations used for short-term projects
such as small departmental applications and rapid development platforms will show
marked cost reductions compared with implementations within the IT department. This
advantages reinforced by the ability to set up a cloud DBMS environment without the use
of expensive IT personnel. The speed of setup will be a primary driver to rapid
deployment of systems without the usual requirements and planning necessary for IT
projects within the IT department. This will also reduce the necessity for IT to respond to
short notice and short duration projects, reducing overall costs in IT. Data management
applications are potential candidates for deployment in the cloud. This is because an on
premises enterprise database system typically comes with a large, sometimes prohibitive
up-front cost, both in hardware and in software. For many companies (especially for startups
and medium-sized businesses), the pay as- you-go cloud computing model, along
with having someone else worrying about maintaining the hardware, is very attractive.
Due to the ever-increasing need for more analysis over more data in today’s corporate
world, along with an architectural match in currently available deployment options, we
conclude that read-mostly analytical data management applications are better suited for
deployment in the cloud than transactional data management applications. We thus outline
a research agenda for large scale data analysis in the cloud, showing why currently
available systems are not ideally-suited for cloud deployment, and arguing that there is a
need for a newly designed DBMS, architected specifically for cloud computing platforms